AEMC’s Proposed Energy Obligations for Data Centres

Australia’s data centre boom raises a difficult policy question: how do we bring large new loads online without materially increasing costs for existing consumers or unnecessarily increasing emissions? The AEMC’s advice focuses on newly connected data centres and accepts that some net increase in system emissions may occur, but seeks to minimise it through four pillars: additional renewables, firming capacity, operational demand flexibility, and efficient network location. The recommendations are primarily designed for the NEM, with the SWIS and Darwin-Katherine systems to be considered separately. It is less clear how large remote or isolated-grid data centres would ultimately be treated.

1. Invest in Additional Renewables

Data centres would be required to procure REGOs associated with “new” renewable generation, with compliance possible through REGOs be they unbundled, via PPAs and on- or near-site generation. REGOs from “recycled” ESEM projects could qualify, meaning projects initially government-underwritten but later supported through commercial arrangements. The proposal also appears to recommend allowing inter-jurisdictional procurement across NEM regions but the wording is open to interpretation. Interestingly, especially given the loose vintage rules on Australian REGOs, REGO liabilities would be split between peak, shoulder and off-peak periods: a middle ground between annual and hourly matching. Transitional arrangements would allow REGOs from existing projects while new projects come online. That seems reasonable; allowing shortfalls to be “made good” using future REGOs is less convincing and risks weakening genuine additionality.

Renewables Open Questions

1. How will “new” renewable generation be defined?
2. Are peak/shoulder/off-peak REGOs reconciled daily, monthly, quarterly or annually?
3. How will tighter data-centre matching coexist with the broader REGO system’s much longer certificate vintage?
4. Must REGOs come from the state of consumption, anywhere in the NEM, or potentially anywhere in Australia?

2. Invest in Firming Capacity

The second pillar introduces a data centre firming contract obligation, requiring operators to cover a proportion of load with firm contracts or qualifying physical capacity. Importantly, the framework is technology-neutral. Compliance could combine behind-the-meter solar and BESS, demand flexibility, long-duration storage or flexible generation with off-site cap contracts, battery tolling agreements, or firmer renewable contracts such as indexed swaps.

This gives operators considerable flexibility to combine firmed and unfirmed renewable procurement. Emergency backup generators would not automatically qualify unless they are available to flexibility markets. A proportion of the obligation would also need to come from new generation and/or firming assets, retaining an additional investment signal.

Firming Open Questions

1. Will firming obligations be tiered according to data-centre size?
2. Could major new demand for cap contracts increase hedging costs for existing industrial consumers?
3. Will the obligation stimulate batteries and longer-duration storage, or inadvertently increase gas generation?
4. What mix of caps, renewable PPAs and BESS tolling ultimately proves cheapest?

3. Provide Operational Demand Flexibility

The third pillar introduces a market-registration obligation, improving AEMO’s visibility over large data-centre loads while creating the regulatory hook for the firming requirements. Registration could also enable suitable data centres to operate as Voluntarily Scheduled Resources (VSRs) and participate more actively in demand response. The critical issue will be economics: the value of flexibility payments must be sufficient to compensate operators for reducing compute availability and potentially affecting customer service commitments.

Operational Flexibility Open Questions

1. Will VSR incentives be strong enough to make participation commercially attractive rather than simply another administrative obligation?
2. If voluntary participation proves insufficient, should minimum demand-flexibility requirements ultimately be imposed on the largest data centres?

4. Locate Efficiently Across networks

The fourth pillar seeks to influence where and how data centres connect. Operators could receive faster or cheaper connections by locating where renewable curtailment is high, co-locating with new firming resources, or offering meaningful demand flexibility. The objective is to encourage data centres to absorb surplus generation, avoid constrained areas and potentially support investment in network stability. Co-location with BESS is particularly interesting: the battery could firm the data-centre load, manage connection constraints and provide power-quality support, while enabling a larger or faster connection than might otherwise be available.

Efficient Location Open Questions

1. Will this pillar complement the other obligations or simply add complexity?
2. Can it genuinely reward low-carbon, grid-supportive projects rather than facilitate gas-backed mega-centres?
3. And where gas provides transitional firming, will the framework create sufficiently strong incentives to replace it quickly with renewables and storage?

Getting Australia’s Data Centre Energy Policy Right

This is a well-intentioned and well-considered first attempt at a difficult policy problem, but the devil will be in the detail. Australia should aim to unlock major investment in data-centre capacity without imposing undue costs on existing consumers or the environment. The impact will not, and arguably should not, be zero given the flow on benefits of housing part of this AI boom in Australia. The challenge is finding the right balance. Let’s not let perfect get in the way of very good.